Chapter 5 Least-cost paths
In this chapter we use the ‘least cost paths’ (a.k.a. LCP) methodology defined in Chapter 3 to define the path of least resistance from all coastal protected areas (larger than 1km2) in the focal region to their corresponding high elevation protected area. For each protected area (Section 1) and mangrove (Section 2), we calculate the LCP for the 10 nearest (by euclidean distance) high elevation protected areas, then select the path with the least resistance. We then explore several important properties of these paths between different countries (e.g. path length, mean resistance, total resistance, quality of forest on the path), in order to understand the spatial state of low-to-high elevation connectivity in central America. Finally, we devise a prioritization scheme to select candidate locations for habitat restoration initiatives.
It is important to note that the LCP method does not fully characterize connectivity in a probabilistic fashion, but it is a tool which cn give us simple information to design prioritization schemes. For a more complete consideration of connectivity, see the next chapter.
5.1 ‘End’ Nodes
We first start by defining a suite of end points (or nodes) - high elevation protected areas. To do this we import all of the high elevation protected area polygons defined in the data assembly section. As the LCP methods require points as targets, not polygons, we also randomly sample within the polygons to create a complete suite of “end nodes”. This means that there are multiple end points within larger protected areas.
In the map below, all of the end points are shown as black points, and their corresponding protected area polygons in blue:
In total, we have a suite (n = 620) of end points for the analyses. For each protected area, we find the 10 closest of each of these end points (euclidian distance) and calculate the least cost paths to them.
5.2 Protected area climate corridors
We used the package ‘gDistance’ and the function ‘shortestPath’ to define the least-cost path from coastal protected areas larger than 1km2 in size to the ten nearest high elevation protected areas. Note, we use starting points adjacent to the ocean to fully encapsulate the ridge-to-reef gradients.
The map below shows the shortest (by distance) path (back lines) for each of the coastal protected areas. The underlying raster layer shows the cost surface, with red areas being highly resistant to movement, and blue areas offering low resistance.
5.2.1 Least-cost-path characteristics
In the following section we explore a suite of characteristics which would make an area a “good” candidate for restoration to preserve low to high elevation connectivity. The categories are as follows:
large starting area: a sufficiently large starting area would ensure strong “supply” of animals, as larger areas have higher wildlife carrying capacities.
large end area: a sufficiently large end location would ensure animals can actually persist in the location they arrive in again, as the carrying capacity for wildlife should be higher.
Short least cost path: short paths will be cheaper to restore than long paths, both in terms of the amount of restoration required, but also in terms of the logistics of tree planting.
Path which is already partially protected: in theory, if a path is fully protected from the start to the end, it will not need any investment in restoration. This might occur if the low elevation areas and high elevation target are within the same protected area. If this is the case, the elevation corridor should be safe from future development.
Path is already of good quality: it may be the case that the path travels through unprotected areas which are high quality forests, these paths would be a high priority for future conservation and resoration initatives.
-Path which doesnt cross multiple administrative borders: the more administrative boudaries (e.g. countries) a path traverses, the more logistically complex its protection and restoration will be.
In the following section, we compare the LCP characteristics between different countries to get a general understanding of the state of ridge-to-reef connectivity.
5.2.1.1 Number of coastal protected areas available
Understanding how many different potential source populations there re per country can give an idea of where might be appropriate to start conservation initiatives. The barplot below shows the number of coastal protected areas (>1km2) by focal country. As you can see, Costa Rica has the largest number of coastal protected areas, followed by El Salvador then Panama. Belize has the least.
5.2.1.2 Starting protected area size
Not all coastal protected areas are equal however, and some have the potential be be quite small. The plot below shows the distribution of start point protected areas by country.
On average, Nicaragua has the largest coastal protected areas. Excluding the marine protected areas, the most substantial are Darien (Panama) and the Río Plátano Biosphere Reserve in Honduras.
5.2.1.3 End protected area size
Having large, protected, end point make the climate corridors more resilient to climate change. The graph below plots the area of end points by country. Panama’s Darien and Costa Rica’s La Amistad are the largest, high elevation, contiguous protected areas in central america. El Salvador, Guatemala and Belize lack large high elevation protected areas.
5.2.1.4 Multinational connectivity
The plot below shows the number of jurisdictional boundarys the climate corridors pass through. Most start and end within the same country (~1). The only exception is Belize, which has no high elevation protected areas of its own. Thus, the paths typically cross into a neighboring country.
5.2.1.5 Least cost path lengths
Short climate corridors will cost less to restore and maintain, longer corridors may be highly logistically complicated. The graph below shows the average LCP length, from start to finish for each corridor. Given the lack of high elevation areas within its borders, Belize has the longest corridors. Costa Rica has, on average, the shortest although Guatemala, Honduras and Panama are similar.
5.2.1.6 Average conductance
Paths we on average, less resistant to animal movement will be more effect as they should be easier to move through. The plot below shows the breakdown by country. Panama appears to be the least disturbed (most conductive) landscape, whereas the habitats in El Salvador, on average, traverse some highly unsuitable habitats.
5.2.1.7 Least-suitable habitat
It may be the case that the utility of a corridor is defined by the worst habitat within it. For example, the habitat is so poor that animals may not traverse it. We explored the minimum conductance of each climate corridor across the landscapes in the plot below. As expected, El Salvador performed very poorly owing to the landscape being highly human modified. Panamas worst scores were typiucally better than the average scores of most other countries.
5.2.1.8 Proportion of the path which is already protected
Understanding the proportion of these climate corridors which are fully protected can give insight into how resilient they are in the future. The plot below shows the proportion of each of the corridors designated as protected by country. Remarkably only 3 of the putative corridors are fully protected to date. These are corridors which terminate in Parque Nacional Cerro Hoya, Panama and Nombre de Dios, Honduras. Panama has several other corridors which are highly protected - including Darien and Reverendo Padre Jesús Héctor Gallego Herrera. Costa Rica’s highest protected climate corridor terminates in the Area de Conservación Guanacaste. As expected, El Salvador, on average, has the least prtected climate corridors. Over 20 putative corrdiors have <10% protection.
5.2.1.9 Proportion of forest cover
Regardless of whether the climate corridors are protected or not, they may have a reasonable amount of forest cover. The plot below shows the average forest cover per climate corridor by country. Panama has the highest average climate corridor forest cover (~90%), followed by Nicaragua. As expected, El Salvador had the lowest average forest cover (~60%).
5.2.2 Multivariate corridor prioritization
Given that there are multiple elements which must be balanced in order to prioritise locations for climate corridors (see above), one approach to integrating them is through the use of a principal component analysis (PCA). PCA’stake multiple input variables and decompose the correlated variation between them into “principal components” - which are essentially a mix of all the variables which load onto them. This allows users to simplify mutiple variables of interest into two or three interprtable axes.
Below, we perform this analysis on the characteristics provided above. Ccorridor length, start area, proportion forested, end area, average habitat conductance, and proportion of protected forest and their loading directions are shown as black lines, individual corridors are shown as points colored by their terminal country.
The graph shows that axis 1 is principally associated with increasing starting area size, forest quality and proportions protected as you move fromn left to right. Axis two is principally associated with path length, with climate corridors increasing in length from bottom to top.
5.2.2.1 The ‘sweet spot’
A sweet spot exists in the above plot which represents climate corridors which:
- are not fully protected, but also have some protection
- have large start and end locations
- are which are not to long
- are already situated in good habitat
These criteria all represent locations where climate corridor restoration initiatives would be suitable and, crucially, achievable. Using a green polygon below, we highlight the suite of corridor which fit this description:
Below we plot these 35 climate on a map to see if we can find a subset of wider landscapes which represent tartgets for climate corridor restoration.
5.2.3 Conclusions
As you can see form the obove map, the selected climate corridors are principally located in Panama, Costa Rica and Honduras. There is not much potential for restoration of climate corridors in El Salvador, and to a lesser extent Nicaragua, Guatamal and Belize.
The majority of corridors lie in and around La Amistad protected area, from both the Caribbean and Pacific directions, highlighting the importance of these regions.
Panama is relatively under represented as it already possess climate corridors with a cool percentage of protection and which contain good forest. These should be considered targets for preservation, not restoration per se.
5.3 Mangrove climate corridors
We now repeat this process for mangrove fragments, to see if there are locations where there may be co-benefits for mangroves and terrestrial protected areas:
The graph below shows the LCPs (black lines) for all mangroves larger than 1km2 in size to high elevation protected areas (blue polygons):
5.3.1 Country characteristics
As above, we will summarize the linkage between mangrove and high elevation protected areas
5.3.1.2 Starting mangrove patch size
Mangroves are typically small ~3km2, but they can be up to 130km2.
5.3.1.5 Least cost path lengths
Mangroves in Costa Rica have, on average, the shortest links to high elevation protected areas. Belize the longest.
5.3.1.6 Average conductance
What is the range in average conductances? Remember, higher = better.
El salvador and Guatemala in the worst shape. Least spread in Costa Rica.
5.3.2 Multivariate summary
As before (see above), we implement a PCA to reduce the suite of important characteristics into two principal axes of variation.
5.3.2.1 The ‘sweet spot’
A sweet spot should exist in the above plot:
- paths which are not fully protected, but also have some protection
- paths with large start and end locations
- paths which are not to long
- paths already situation in good habitat
Lets plot these areas 151 on a map a see if we can identify 10 reasonable candidates areas for restoration:
5.4 Comparing protected area and mangrove climate corridors
Areas where climate corridors for mangroves and protected areas overlap will provide co-benefits for habitat restoration and may help in the prioritization process. Below we map both sets of putative corridors, protected areas = “red” and Mangroves = “blue”, to determine if there is any evidence for overlap: